Innovative Statistical modelling for a better Understanding of Longitudinal multivariate responses in relation to Omic datasets

ended

In medicine and agronomy, there is a growing interest in identifying biological mechanisms involved in the evolution of traits along time. Nowadays, this challenging objective is achieved through the acquisition of high-dimensional –omic datasets from various biological levels, and with the collection of phenotype measures along time on the same individuals, so-called longitudinal data. A new research focus is emerging with the objective to analyze jointly these two types of data. In this project, we propose to develop innovative statistical methods to simultaneously analyze these types of data and to deal with their respective characteristics. Novel methodologies will be developed by combining statistical concepts from linear mixed model and variable selection in a Bayesian framework, and by incorporating or inferring biological relationships. The first aim will focus on the analysis of one or more longitudinal outcomes with one –omic data. Flexible modeling for approximating time-varying covariate effects combined with variable selection approaches will be proposed. Thus, a better understanding of the relationships along time between the outcomes and the relevant covariates will be reached. The second objective is to investigate the integration of multiple –omic datasets for explaining one univariate outcome, then one longitudinal response variable, and finally multivariate longitudinal outcomes. Bayesian hierarchical modeling with prior distributions allowing to capture relationships among –omic datasets will be investigated and new relationships among –omic datasets will be explored. The developments and findings of this project research will greatly contribute to the statistical and biological domains. New generic statistical methods will be developed and will be available for transversal applications in various fields. Finally, this project will highlight the added value brought by a collaborative and interdisciplinary work with experienced researchers.